Stochastically Perturbed Parmaters [SPP]

What parameters do we perturb?

• Cloud thickness threshold for shallow / deep convection
• Cloud ice content impact on cloud thickness
• Ice nucleii concentration
• Saturation limit sensitivity for condensation*
• Kogan autoconversion speed
• Kogan sub grid scale (cloud fraction) sensitivity*
• Impact of graupel on radiation
• Impact of snow on radiation*
• Top entrainment efficiency
• Stable conditions length scale
• Asymptotic free atmosphere length scale

However… This Presentation is not as advertised!

• Original reference runs had problems with warm perturbed members in winter and dry perturbed members in summer
• Original reference did not include modelling of snow on sea ice
• Some default physics parameter values are changed for SPP

• The reference is re-run with
  • 6 perturbed members
  • No surface data assimilation for perturbed members
  • Snow on sea ice modelling
  • Physics parameter defaults changed

Point verification for (near) surface parameters

2m Temperature

2m Temperature [members]

2m Temperature “first guess”

Temperature [upper air]

2m Humidity

2m Humidity [members]

Relative Humidity [upper air]

10m Wind Speed

10m Wind Speed [members]

Wind Speed [upper air]

6h Precipitation

6h Precipitation [members]

Total Cloud Cover

Total Cloud Cover [members]

Geopotential Height [upper air]

Summary

• Warming of perturbed members reduced
• Night time wet bias removed in perturbed members
• Large positive bias in initial cloud cover for perturbed members removed
• Overall reduction in spread for all parameters
• Less spread throughout atmosphere